Characterization of responses to temperature and photoperiod for time to flowering in a world lentil collection.

The times from sowing to first flowering (f) of 231 accessions of lentil (Lens culinaris Medik.), comprising germ plasm from eight countries and breeding lines from ICARDA in Syria, were recorded in four glasshouse environments; two photoperiods (16 and 13 h/day) combined with warmer (24°/13°C) and cooler (18°/9°C) day/night temperatures. The linear model 1/f=a+bT + cP (where T is mean diurnal temperature and P is photoperiod) provided an average fit over the 231 accessions of r (2)=0.852. Since there is no interaction term in this linear model, the flowering responses of an accession to temperature and photoperiod are independent. The values of the constants b and c indicate relative responsiveness of rate of progress towards flowering (1/f) to temperature and photoperiod, respectively. Comparison among the 231 accessions showed a weak, but significant, negative correlation between the values of b and c (r=-0.291, P<0.01). Since the proportion of the variance of b not attributed to its linear regression on c was >0.91, we conclude that these phenological responses are under separate control and that there is considerable scope for selection of any combination of sensitivities to temperature and photoperiod in lentil. Just as a large proportion of the variation among accessions in mean time to first flowering was attributed to country of origin, so also was variability in the values of the constants a, b, and c. In particular, sensitivity to photoperiod (i.e., the value of constant c) was dependent upon latitude of origin. Breeding lines from ICARDA were equally variable in a, b, and c as were germ plasm accessions from elsewhere, while the mean values were similar to those of accessions from neighboring Jordan. A single accession of wild lentil (L. culinaris subsp. orientalis) from Turkey showed flowering responses to T and P similar to the mean value of accessions of cultivated lentil from that country. Results from diverse environments for the Argentinian cv Precoz show that the use of this linear model facilitates predictions of time to flowering in any environment (within wide limits) of known mean temperature and photoperiod. The model, then, minimizes the need for multisite evaluations of phenology, since predictions of pre-flowering duration in any environment, and characterization of flowering responses to photoperiod and temperature, can now be achieved by screening germ plasm in a few, carefully selected locations.